00:00:04
so the researchers really looking at a
00:00:07
puzzle that we observed um in terms of
00:00:10
new technologies being introduced into
00:00:12
the market but significant differences
00:00:14
in terms of how fast they're able to
00:00:17
reach mainstream adoption and disrupt
00:00:20
existing markets existing players so we
00:00:22
we observe that in the printer space
00:00:25
injet printers came about quickly took
00:00:28
over the dot matrix we look at stdv it
00:00:31
took decades for it to be mainstream
00:00:33
adoption and then we look at things like
00:00:35
the segue or the Palm uh types of
00:00:39
Technologies which either created some
00:00:42
value or never really reached mainstream
00:00:44
adoption so so the question was really
00:00:46
what explains that why some technologies
00:00:49
are introduced and immediately supplant
00:00:51
existing Technologies whereas others
00:00:53
take decades or sometimes don't really
00:00:56
reach mainstream adoption and so what we
00:00:59
did was we we we tried to find a
00:01:02
context where um we could observe enough
00:01:05
variation in terms of how quick or slow
00:01:09
these technology adoption patterns were
00:01:11
but helped us to control for a lot of
00:01:13
confounding effects as well and so the
00:01:16
idea was to find a natural experiment so
00:01:18
to speak and uh that helps us to ensure
00:01:22
that the source of variation is not
00:01:24
driven by some systematic effects which
00:01:26
may make inferences more problematic so
00:01:29
what we did was we we came across you
00:01:32
know I had some experience in the
00:01:33
semiconductor industry we came across
00:01:35
the semiconductor lithography technology
00:01:38
as as a setting that we could study this
00:01:41
question um what is interesting about
00:01:43
this setting is this is kind of the
00:01:44
engine behind more slow or any progress
00:01:48
in semiconductors over the last 40 years
00:01:50
has been fueled by lithography
00:01:52
technology so it's it's important we
00:01:55
know it's fast-paced and and we thought
00:01:58
that this is where we may want to study
00:02:00
this question and so the research design
00:02:02
looked at the semicon lithography and 10
00:02:05
different new technologies that were
00:02:07
introduced in that industry over a
00:02:09
40-year period um you know we
00:02:11
interviewed about 30 industry experts to
00:02:13
try to get a sense for what's driving
00:02:16
these patterns we collected data on
00:02:18
Technologies markets uh
00:02:21
Industries um in terms of really trying
00:02:23
to understand the factors that may be
00:02:25
explaining this difference and uh what
00:02:28
we found was was interesting you know a
00:02:30
lot of the research and practice focuses
00:02:33
on on on new technologies and how it
00:02:36
interacts with the markets and look at
00:02:38
is the technology better than what's
00:02:40
available now or it's not and that
00:02:42
explains whether it's going to reach
00:02:44
mainstream adoption what we found was
00:02:46
that's sort of part of the explanation
00:02:48
and actually in our case a very small
00:02:50
part of the explanation we found that
00:02:52
the bigger explanation was not focusing
00:02:55
on the new technology and the market but
00:02:58
looking at the technology ecosystem
00:03:01
which is to means that how is the
00:03:02
technology created what are the
00:03:04
different elements that make up the
00:03:06
technology so think about batteries
00:03:07
making electric cars but also how is the
00:03:10
technology used by the users in terms of
00:03:13
other elements or complimentary
00:03:15
Technologies and services think about
00:03:17
electric car and charging stations and
00:03:19
garages who can fix that electric cars
00:03:21
and and not only that uh that you need
00:03:24
to look at the technology ecosystem for
00:03:26
both the new technology and the old
00:03:29
technology
00:03:30
so what we set out is you know we want
00:03:32
to explain this
00:03:33
variance what we see from our our our
00:03:36
fieldwork and our interviews that by
00:03:39
looking at the technology itself the
00:03:41
resolution in terms of fast versus slow
00:03:43
is not going to be very good let's look
00:03:46
at the ecosystem and so as as part of
00:03:49
the research we collected data that
00:03:51
systematically identifies each
00:03:53
technology in terms of its ecosystem and
00:03:56
and compared in terms of the existing
00:03:58
technology and what we found was because
00:04:00
when we think about the new technology
00:04:02
there's sometimes what we call an
00:04:03
ecosystem emergence challenge which
00:04:06
means the technology is ready but the
00:04:08
ecosystem still needs some Investments
00:04:10
like the charging infrastructure for
00:04:12
electric car for it to reach mainstream
00:04:14
adoption and we call that ecosystem
00:04:16
emergence challenge but we also see in
00:04:18
the old technology sometimes you can
00:04:20
extend that technology by improvements
00:04:22
in components or these complimentary
00:04:24
elements so think about the gasoline
00:04:26
cars you know nobody thought 10 15 years
00:04:28
ago that they will be going at 30 m per
00:04:30
gallon or even 40 m per gallon but
00:04:32
improvements in engines improvements in
00:04:34
fuels allowed the cars to get to that
00:04:37
point you think about the hybrid versus
00:04:39
electric car uh sort of discussion where
00:04:41
the hybrid cars introduced were able to
00:04:44
grow market share much faster than
00:04:46
electric cars and the main difference is
00:04:48
the hybrid cars could plug and play
00:04:50
there wasn't an emergence challenge in
00:04:52
the ecosystem electric cars had these
00:04:55
emergence challenges by creating the
00:04:57
infrastructure by having these charging
00:04:59
stations in place so at the end of the
00:05:01
day after doing this research we were
00:05:04
able to document fairly clearly that if
00:05:06
one were to understand the likelihood
00:05:09
that the new technology is going to come
00:05:10
and immediately disrupt the marketplace
00:05:13
versus it may take much longer or may
00:05:15
not happen you have to look at the new
00:05:17
technology ecosystem in terms of its
00:05:19
emergence Challenge and the old
00:05:21
technology ecosystem in terms of the
00:05:23
extension opportunity and it's really
00:05:25
The Joint consideration of these two
00:05:28
factors that explain
00:05:30
whether you will see a fast technology
00:05:32
takeoff or a technology that will never
00:05:34
reach mainstream
00:05:39
adoption you know the research presents
00:05:41
some very interesting takeaways for
00:05:44
managers for uh policy makers for
00:05:48
investors into technology companies and
00:05:50
also for users of Technologies whether
00:05:52
you're consumers or businesses so if you
00:05:55
are a manager of a firm whether it's a
00:05:57
new startup or it's an established firm
00:06:00
you always have to think about sort of
00:06:02
resources that you have to allocate
00:06:03
towards new technologies and how you
00:06:06
transition between an existing to a new
00:06:08
technology so think about Kodak shifted
00:06:10
from the chemical based to the digital
00:06:12
photography or you think about Netflix
00:06:15
moving from a DVD rental business to the
00:06:18
online streaming business and so what we
00:06:20
find is you know managers can use this
00:06:23
framework to set realistic expectation
00:06:25
in terms of both whether and when to
00:06:29
invest in new technologies and it may
00:06:32
sometimes actually make sense and they
00:06:34
will generate more shareholder value or
00:06:35
destroy less shareholder value uh by
00:06:38
focusing on the existing Technologies
00:06:40
than kind of going all out and pushing
00:06:43
for the new technology from an
00:06:44
Investor's perspective again it sets
00:06:46
realistic expectations uh when you are
00:06:48
investing in technology companies in the
00:06:51
ability to create value over time you
00:06:53
know sometimes the expectations could be
00:06:54
2 years but as we know anecdotally it
00:06:57
often takes much longer than that for
00:06:59
new technologies to create value users
00:07:02
you know same same sort of decisions
00:07:04
whether you're a business or a consumer
00:07:06
you're always up against the decision
00:07:07
should we go for the latest and greatest
00:07:10
or should we wait until until the new
00:07:13
technology is developed to a point where
00:07:15
it creates value and so having the
00:07:17
sensibility in terms of existing
00:07:18
ecosystem and new ecosystem can help
00:07:20
them make more optimal decisions policy
00:07:23
makers I mean this is a big question in
00:07:25
terms of the role of policy in shaping
00:07:27
technological progress
00:07:29
and we know that a lot of economic
00:07:31
progress that we see is driven by
00:07:33
improvements in technology so if you run
00:07:35
into a scenario where the new
00:07:37
technologies are not emerging at the
00:07:40
rate where you expect them to to to to
00:07:42
sort of emerge you know that has a
00:07:44
downside in terms of jobs in terms of
00:07:46
economic output in terms of GDP growth
00:07:49
so thinking more in terms of ecosystem
00:07:52
both for the new technology and the
00:07:54
existing technology helps all of these
00:07:56
actors get better Returns on the
00:07:58
investment
00:07:59
get less surprises in terms of
00:08:05
expectations you know that was the
00:08:07
original intention in terms of our
00:08:10
ability to
00:08:11
forecast uh how Technologies are going
00:08:13
to evolve over time and the title of the
00:08:16
paper has S curves as a way people have
00:08:18
thought about it and there two different
00:08:20
ways people thought about forecasting
00:08:22
Technologies one is thinking about the
00:08:24
performance improvements of the
00:08:26
Technologies and we know that those
00:08:27
performance trajectories tend to be in s
00:08:29
shaped early stage you invest a lot but
00:08:32
you're not getting improvements then
00:08:33
there's a takeoff where you expect most
00:08:36
of the market takeoff and then there's
00:08:37
maturity similarly like that there's an
00:08:39
adoption esurf which is the early stage
00:08:42
of the technology the users who are
00:08:43
going to be going after it are not the
00:08:45
ones who really care about the total
00:08:46
value but they just like the new
00:08:48
technology that's a very small part of
00:08:50
the market the mainstream users are
00:08:52
looking to see what's the main value
00:08:54
proposition not just technology being
00:08:56
new and that depicts an s-shaped
00:08:58
distribution as well so what we what we
00:09:01
were up against is these existing
00:09:04
Frameworks and tools which has this
00:09:06
nonlinear s-shaped Dynamics and what we
00:09:10
are able to show through this ecosystem
00:09:12
based sensibility is that the pattern of
00:09:15
these Evolutions both in terms of
00:09:17
Technology Improvement and market
00:09:18
adoption may not fit well with a smooth
00:09:22
s shaped trajectories and they often
00:09:24
could be very nonlinear patterns could
00:09:26
be very discontinuous patterns and
00:09:29
having the sensibility can help you
00:09:30
predict right whether the S shapes are
00:09:33
going to be smooth and quick or these
00:09:36
may not be s shaped and they may
00:09:37
actually get resolved in a much longer
00:09:40
time frame and and you know where I'm
00:09:42
going with this project is exactly along
00:09:44
those lines is to have um a a model and
00:09:48
an approach that helps us to improve
00:09:50
technology
00:09:51
forecasting in fact one of the things I
00:09:53
was surprised um when we were doing this
00:09:56
research is um the amount of resources
00:10:00
that both companies and governments were
00:10:02
expending into these new technologies
00:10:05
often running into hundreds and billions
00:10:06
of dollars as you know semiconductor
00:10:08
industry is very Capital intensive and
00:10:11
despite these resources and expectations
00:10:13
they have well documented industry road
00:10:15
maps that say these new technologies are
00:10:17
going to be hitting mainstream 3 to 5
00:10:19
years more often than not they would
00:10:21
take double the amount of time or in
00:10:24
many cases never take off at all um and
00:10:28
that's something that I was not some not
00:10:31
expecting as much I was expecting some
00:10:33
of it but not at the scale that I
00:10:34
observe the other issue that I thought
00:10:36
was particularly surprising is you know
00:10:39
we kind of write off existing
00:10:41
Technologies and old Technologies you
00:10:43
know the the cell phones nobody's going
00:10:45
to use it everybody's going to shift to
00:10:46
let's say tablets on and a newer newer
00:10:49
ways to do things um but seemingly
00:10:52
geriatric
00:10:53
Technologies uh continue to create a lot
00:10:56
of value for a very long time and that
00:10:59
was counterintuitive for what we
00:11:01
expected to see in this result
00:11:02
especially in an industry which is being
00:11:04
driven by Moors
00:11:09
law so I think from a firm perspective
00:11:11
as we said you know the big implication
00:11:13
is thinking about resource allocation
00:11:15
and whether it makes sense for firms to
00:11:17
go all out on these new technologies at
00:11:19
the rate that they expect to invest um
00:11:21
the other implication from a firm
00:11:23
perspective is timing um you know in a
00:11:26
related study I was able to document
00:11:28
that you know first mover Advantage as
00:11:30
we often claim in many technology
00:11:32
settings turns out not to be the case
00:11:35
and in fact we were able to show
00:11:36
significant first mover disadvantage in
00:11:39
Technologies where the ecosystem
00:11:41
emergence challenge was so high that
00:11:44
first movers couldn't create any value
00:11:46
so issues of timing was important
00:11:49
another issue that I think uh you know
00:11:51
managers and uh investors need to think
00:11:53
about is a technology is seldom a single
00:11:56
artifact or a single technology element
00:11:59
it is the ecosystem and that makes it
00:12:02
very difficult for firms to control all
00:12:04
the elements that go into creating value
00:12:07
from the technology so one sensibility
00:12:10
that you know I'd like to share with uh
00:12:12
the managers and and investors is as we
00:12:14
think about evaluating technological
00:12:16
opportunities it's not enough to look at
00:12:19
just the focal technology whether it's
00:12:20
the phone or it's a computer or it's the
00:12:22
car is to think about the ecosystem what
00:12:25
are the elements that go into it and
00:12:27
could you orchestrate the ecosystem
00:12:28
system in a way that mitigates these
00:12:31
potential downsides of the users not
00:12:34
deriving value when the technology is
00:12:36
ready um you know what was uh
00:12:38
particularly interesting from our study
00:12:40
was every time the new technology was
00:12:42
introduced that technology was actually
00:12:44
Superior in terms of performance but
00:12:47
despite that it never reached mainstream
00:12:49
adoption and the explanation was really
00:12:52
rooted in the ecosystem of the new
00:12:53
technology and the ecosystem of the old
00:12:56
technology we had a we had a case where
00:12:58
new techn ology came in three times four
00:13:00
times Superior performance but a
00:13:03
critical element that the users need to
00:13:05
use so think about you have a camera as
00:13:07
this technology but you need the film
00:13:09
but the film that is you available with
00:13:11
this new technology is not as good for
00:13:14
you to get the full potential of this
00:13:17
new camera so users don't really see an
00:13:19
incentive so you're coming up with the
00:13:21
best camera but with a below par film
00:13:24
and the total value proposition is not
00:13:26
there in other cases the Technologies
00:13:28
are fairly increment m al but they're
00:13:29
able to Plug and Play like a hybrid car
00:13:32
for example and that doesn't feel that
00:13:35
doesn't see these sort of resistances
00:13:37
and fictions in the marketplace uh from
00:13:39
a venture capital perspective again as
00:13:41
we think about these entrepreneurial
00:13:43
ecosystems as a way for Venture capitals
00:13:45
to to sort of get in and create value um
00:13:49
again you know you cannot just localize
00:13:51
on a single firm as a basis of
00:13:53
investment as a basis of value creation
00:13:55
if you think about ebooks as a case in
00:13:57
point U you know a lot lot of venture
00:13:59
capital Investments companies like e Inc
00:14:01
back in the day were funded through a
00:14:03
lot of VC money both uh private Venture
00:14:06
Capital but also corporate Venture
00:14:08
Capital um but it wasn't about the ink
00:14:11
itself it was about content it was about
00:14:13
the reader it was about Amazon's
00:14:15
business model that had to bring the
00:14:17
solution together so as we think about
00:14:18
the opportunities for venture capital
00:14:20
for any given technology it's not enough
00:14:22
to just focus on that technology but
00:14:24
thinking about the broader ecosystem and
00:14:26
how one could control and dve
00:14:29
progress through the
00:14:34
ecosystem important uh misperception
00:14:37
that you see quite often is you know
00:14:39
fairly optimistic aggressive
00:14:42
expectations about new technologies and
00:14:45
some people call it Hypes and some
00:14:46
people you know refer to it as hype
00:14:48
Cycles um and I think we're able to kind
00:14:51
of identify
00:14:52
reasons of those Hypes and just calling
00:14:55
them as a hype is not enough in in my
00:14:58
view is to understanding the factors
00:15:00
that drive the hype and then you know
00:15:02
using that as a basis of making good
00:15:05
decisions um so you know there was
00:15:07
clearly very high expectations in in our
00:15:09
research context that these Technologies
00:15:12
would reach mainstream adoption we going
00:15:14
to be investing Millions if not billions
00:15:16
of dollars into into these Technologies
00:15:19
but it turned out those Investments were
00:15:21
in vain because either the ecosystem
00:15:23
didn't emerge at the right time or at
00:15:25
the cost that made it attractive for the
00:15:27
users
00:15:29
the second sort of uh misperception is
00:15:31
again this um you know very dramatic
00:15:35
views about the decline of the old
00:15:38
Technologies and this you know framing
00:15:40
of the world as a world of disruption
00:15:42
and all the values going to be created
00:15:44
through creative destruction coming from
00:15:46
new technologies I you know we do feel
00:15:48
that that's that part of the world is a
00:15:51
bit oversold and in fact uh a lot of the
00:15:54
value uh in semiconductors in many of
00:15:57
the other industries that I've studied
00:15:59
is is been created through continued
00:16:01
Innovations and finding Market
00:16:03
opportunities through existing
00:16:04
Technologies and we found that in our
00:16:06
context as well a very interesting
00:16:08
company called ultratech has been in
00:16:10
this industry for 40 years that we have
00:16:12
studied this industry it has never been
00:16:14
at The Cutting Edge of the technology
00:16:16
like some of its peers who entered and
00:16:18
exited in a fairly short span and this
00:16:21
company has existed for a very long time
00:16:24
and still creating a lot of shareholder
00:16:26
value so I think I think those
00:16:28
perceptions in terms of you know a very
00:16:31
optimistic view of of the new
00:16:34
technologies and a very pessimistic view
00:16:37
of the value proposition in the old
00:16:38
technology is something that we think
00:16:40
needs to be more balanced as opposed to
00:16:43
these extreme
00:16:47
views so we are certainly not the first
00:16:50
ones to study new technologies and the
00:16:53
ability to create value in the
00:16:54
marketplace um but I do think that we
00:16:57
are one of the first if if not the first
00:16:59
to bring to attention the important role
00:17:02
of the ecosystem in the way the
00:17:04
technology gets developed and
00:17:05
commercialized and to really understand
00:17:08
how these Technologies transition and
00:17:09
Technology Dynamics we think it's not
00:17:11
enough to just take the ecosystem
00:17:13
perspective but look at both the new
00:17:15
technology ecosystem and the old
00:17:17
technology ecosystem it's really
00:17:19
bringing these two ecosystems together
00:17:22
as an analytical framework can generate
00:17:24
a lot more valuable insights in our view
00:17:26
and we showed that in our research
00:17:28
by just looking at the new technology on
00:17:31
its own you know the other sort of
00:17:33
aspect of the research which I think you
00:17:34
know I'm particularly proud of is the
00:17:37
approach that we took to study this
00:17:39
question it was you it required almost
00:17:42
two years of field work we interviewed
00:17:44
more than 30 industry practitioners
00:17:46
managers Consultants from a variety of
00:17:49
roles in the ecosystem and that took a
00:17:52
lot of time but we felt that you know we
00:17:55
were able to generate a set of robust
00:17:57
findings that made that effort
00:18:00
worthwhile something that we don't often
00:18:02
see in a lot of management
00:18:07
research so this was in my view you know
00:18:10
probably a first step for me to try to
00:18:13
think about how one could make better
00:18:15
decisions in terms of new technologies
00:18:18
and provide a framework that could guide
00:18:21
that uh decision- making I think we're
00:18:23
still not quite at the point I would
00:18:26
like us to be in terms of our ability to
00:18:28
forecast Technologies both existing and
00:18:31
new and and to kind of take a more
00:18:33
probabilistic view that can guide
00:18:36
managerial decision- making Venture
00:18:38
Capital Investments and even how policy
00:18:40
makers think about so what I'm I I'm
00:18:42
sort of moving on is to take the issue
00:18:45
of Technology forecasting more seriously
00:18:48
and anecdotal evidence suggests that
00:18:50
we're actually very poor technology
00:18:52
forecasters I mean there's not a lot of
00:18:54
uh documentation in terms of accuracy of
00:18:57
forecasting but if you look at some of
00:18:59
the the big technology forecast that
00:19:01
come out nobody tracks them which is a
00:19:03
problem of course but anecdotally you
00:19:06
can tell that you know a vast majority
00:19:08
of them turn out to be wrong either in
00:19:10
terms of missing the timing completely
00:19:13
or missing the new technologies that
00:19:16
nobody predicted 3 to 5 years ago um so
00:19:19
my goal is to think about the
00:19:21
theoretical logic but also think about
00:19:23
an estimation modeling approach that
00:19:27
helps us to predict new technologies in
00:19:29
a way that we have not able to do so far
00:19:32
and I'm starting with the auto sector
00:19:35
which which is going through some fairly
00:19:37
interesting uh shifts at the moment as
00:19:40
you've shifted from the electrification
00:19:42
now we're talking about autonomous cars
00:19:44
we have the technology companies coming
00:19:46
into it in addition to the auto sector
00:19:48
um and so I think it's it's an
00:19:50
environment that presents a lot of
00:19:52
uncertainty but also a lot of
00:19:53
variability in terms of technological
00:19:56
choices being pursued so I'm working on
00:19:58
a project where uh we will try to create
00:20:02
good forecasting models and hopefully we
00:20:05
can show the accuracy of these
00:20:06
forecasting models in explaining many of
00:20:09
the Technologies being pursued in the
00:20:11
auto sector and the hope is to kind of
00:20:12
take that as a template and apply it in
00:20:15
other technology settings you know think
00:20:18
about internet of things think about
00:20:20
variable Technologies think about
00:20:22
financial Technologies and we could
00:20:24
scale this model to make better
00:20:26
technology forecast
00:20:39
[Music]
